WD Ultrastar incessant 5 second seek

[ptyork]

I have 12 14TB WD drives. 4 are shucked (used for a backup pool). 4 are WD Red Plus. 4 are "new" refurbished Ultrastar HC530's. I've just recently created a 8x RAIDz2 from the Reds & Ultrastars. All are in a DIY case with either a 8x SAS2308 or little 4x SATA card, though the symptoms occur regardless of which the drives are attached. TrueNAS Scale (latest stable).

So the symptoms are a "clack" head seek every 5 seconds but ONLY on the Ultrastars. I can remove each of the 4 drives Ultrastars in sequence and it continues, so I don't think it's just one of them (i.e., likely not a defective drive issue). Though I may not keep it this way, at least for experimentation I've got APM and spin down enabled on all 12, and the other 8 will spin down; Ultrastars not so much. Moving all four between backplanes and between controllers...same deal. Export the pool...clacking stops. There are no running tasks and I don't really think this is actual disk I/O on the pool since the other 4 in the array will spin down while the clacking continues.

I'm at a loss to explain it with my limited knowledge. I don't mind the lack of APM, but the clacking worries me. And drives me insane since it's quite audible in an otherwise near-silent build. Is this just an enterprise drive thing? Is there an easy fix?

Checking here before going to the iX forums since you guys tend to be a bit kinder to the less experienced among us.

 

[ptyork]

Okay, I'll answer my own question here (truly I tried forever to figure it out before posting). Apparently it was the Kubernetes-based "Apps" service (k3s). I had no apps running, but I had set the pool to the big new data pool. Apparently the service was doing something to ping these drives. No idea why ONLY the Ultrastars were getting hit, though perhaps the sectors being read/written were only on these drives? Ah well, problem solved. Insanity averted.

 

[Whaaat]

And drives me insane since it's quite audible in an otherwise near-silent build. Is this just an enterprise drive thing? Is there an easy fix?

That is how background medium scan works in Ultrastars. Loud head movement each 4-5 seconds. And that's why I prefer SAS drives. Disable BMS using sdparm and enjoy the silence.

 

[ptyork]

Disable BMS using sdparm and enjoy the silence.

Interesting. Thanks for the info! I'd assumed that the k3s service was writing every 5 seconds, but perhaps it was just keeping it online and thus triggering the BMS. Hmmmm. Anyway, sdparm didn't work for me, but it looks like I might also be able to do this with smartctl (View topic - How to turn off background media scan?). I'm currently running a long smart scan, which may explain the sdparm failure. Regardless, I'll experiment with this when done.